Post on 15-Aug-2020
Path-Dependent Option Valuation When the Underlying Path
Is Discontinuous
Chunsheng Zhou y
March 1997
yFederal Reserve Board, Mail Stop 91, Washington, DC 20551. Tel: (202) 452-3328, Fax: (202)
452-5296, E-mail: czhou@frb.gov. The author thanks Katherine Allen for research assistance. The
analysis and conclusions set forth in this paper are those of the author and do not indicate concur-
rence by other members of the research sta�, by the Board of Governors, or by the Federal Reserve
Banks.
Abstract
The payo�s of path-dependent options depend not only on the �nal values, but also on
the sample paths of the prices of the underlying assets. A rigorous modeling of the under-
lying asset price processes which can appropriately describe the sample paths is therefore
critical for pricing path-dependent options.
This paper allows for discontinuities in the sample paths of the underlying asset prices by
assuming that these prices follow jump di�usion processes. A general yet tractable approach
is presented to value a variety of path-dependent options with discontinuous processes. The
numerical examples show that ignoring the jump risk may lead to serious biases in path-
dependent option pricing.
Jumps are an important feature of the price processes of �nancial assets, and are espe-
cially pronounced for certain types of assets, such as small-cap stocks and exchange rates.1
For this reason, Merton's (1976) groundbreaking work, which explicitly admits jumps in the
underlying asset prices for pricing standard European options, has generated a profound
impact on the �nance profession.
Merton's work was done a couple of decades ago. In recent years, the variety of new
option contracts has increased enormously. Many types of so-called exotic options are now
popular items in the over-the-counter market. Most exotic options, such as barrier options,
lookbacks, Asian options, American capped options, and many others, are path-dependent,
that is, their payo�s depend not only on the �nal values, but also on the sample paths of the
underlying asset prices. For these options, a rigorous consideration of jumps in the sample
paths of the underlying asset prices seems more important than for the standard non-path-
dependent options in Merton's world. But these new path-dependent options are often so
complicated that they cannot be priced in Merton's framework. For this reason, this paper
presents a new general yet tractable approach to value a variety of path-dependent options
whose underlying prices follow jump-di�usion processes.2
If a jump-di�usion process of the underlying asset price is misspeci�ed as a pure dif-
fusion process, what will happen to the pricing performance of the option pricing model?
For a given option, say a down-and-out call, will it be overpriced, or underpriced by the
model due to the misspeci�cation of the price process? To answer these questions, several
numerical examples are provided on barrier options and lookback options. The examples
show that for the path-dependent options, ignoring the jump risk often leads to serious
biases in the pricing of both short maturity and long maturity options. The examples
also show some interesting and sometimes surprising maturity patterns of pricing errors
1See, Kon (1984), Ball and Torous (1985), Jorion (1988), and Bates (1991, 1996), Das, Foresi and
Sundaram(1996).2In a recent paper, Amin (1993) provided a discrete time approach to value one kind of path dependent
option| an American option under jump di�usion processes. The approach presented here is much di�erent
from Amin's approach.
1
caused by the misspeci�cation. For instance, misspecifying a jump-di�usion process as a
di�usion process will understate the knock-out probability for short-maturity options, but
signi�cantly overstate the knock-out probability for long-maturity options. As a result, this
misspeci�cation often overvalues short maturity down-and-out calls, but underprices most
long maturity down-and-out calls.
The rest of this paper is organized as follows: The next section presents a general
approach to valuing path-dependent options with jump-di�usion processes. Sections 2 and
3 provide numerical examples to show the impact of the jump risk on the prices of barrier
options and lookback options. Section 4 concludes.
1 The Model
This section presents a general jump-di�usion model to value various path-dependent op-
tions. The model is based on the following assumptions.
Assumption 1: The dynamics of the underlying asset price S are governed by the
jump-di�usion process
dS=S = (�� ��)dt+ �dZ + (�� 1)dY (1)
where
�, �, �, and � are positive constants;
Z is a standard Brownian motion;
dY is a Poisson process with intensity parameter �;
� > 0 is the jump amplitude with expected value equal to � + 1, and
dZ, dY , and � are mutually independent.
Because � equals the expected value of the jump component (� � 1), � in the above
equation represents the expected instantaneous rate of change of the asset price S.
We assume that � is an i.i.d. log-normal random variable, such that
ln(�) � N(��; �2�): (2)
2
This assumption implies that
� := E[�� 1] = exp(�� + �2�=2) � 1:
The di�usion process in equation (1) characterizes the \normal" vibration in the asset
price, due to gradual changes in economic conditions or the arrival of new information which
causes marginal changes in the asset value. The jump component describes the \abnormal"
variations in the asset price due to the arrival of important new information.
Assumption 2: The capital asset pricing model (CAPM) holds for equilibrium security
returns and the jump component of the underlying asset price S in equation (1) is purely
asset-speci�c and is uncorrelated with the market.
This paper considers the options of individual stocks. According to Merton (1976),
there generally does not exist a set of portfolio weights that will eliminate the \jump" risk.
A Black-Scholes hedge will not be riskless even in a continuous-time setup. To validate
the Black-Scholes \risk-neutral" argument, some extra restrictions on the economy and the
jump process must be imposed. If the jump component represents nonsystematic risk, a
portfolio which removes the risk of di�usion component (i.e., dZ does not appear in the
return process of the portfolio) will have a zero \beta." By the CAPM, the expected return
on that portfolio must equal the riskless rate. The jump risk will therefore not receive a risk
premium. Assumption 2 has been widely used as a �rst-order approximation for jumps in
the prices of individual stocks, even though it is not a valid assumption for jumps in market
indexes.
With an increase in complexity, Assumption 2 may be replaced by some alternative
assumptions. For example, one can assume that jumps are systematic and that economic
agents have some speci�c preferences so that risk premium for jumps can be determined
in the model. Bates (1991, 1996) �nds that under a traditional assumption of preferences,
the `risk-neutral' movement of the underlying asset price follows a jump-di�usion process
similar to the one under Assumption 2, so the basic approach in this paper is still valid
for handling nonsystematic jumps. In other words, Assumption 2 simpli�es the model, but
does not change the main conclusions of this paper.
3
Denote W (�;St : 0 � t � T ) as the �nal payo� of a path-dependent option contract,
where � is the set of relevant parameters and (St : 0 � t � T ) is the time-path of the price
of the underlying asset. Using a standard risk-neutrality approach, we have
Lemma 1 The option price F (�; T ) is given by
F = exp(�rT )EQ[W (�;St : 0 � t � T )]; (3)
where EQ represents the expectation under the equivalent martingale measure Q conditioning
on information currently available. Under this measure,
dS=S = (r � ��)dt+ �dZ + (�� 1)dY: (4)
Equation (4) can be rewritten as:
d ln(S) = (r � �2=2� ��)dt+ �dZ + ln(�)dY: (5)
This lemma provides a general framework for option valuation. The next two sections
will use this lemma to value barrier options and lookback options.
2 Pricing Barrier Options
Barrier options are path-dependent options which are either activated (knocked-in) or ter-
minated (knocked-out) if the underlying asset value reaches or passes a speci�ed trigger
level, say H, between inception and expiration.3
After activation and before extinction,
barrier options behave identically to standard European-style options. These options have
become increasingly popular in recent years. Less expensive than standard options, they
may provide the appropriate hedge in a number of situations. For example, a down-and-out
call with a low barrier o�ers an inexpensive protection against a big rise in the underlying
asset price.
3For detailed discussions of various barrier options, see, for example, Rubinstein and Reiner (1991), Heyne
and Harry (1994), Boyle and Lau (1994), Ritchken (1995), and Rich (1994).
4
For barrier options, jumps not only a�ect the distribution of the value of the underlying
asset at the maturity of an option, but also in uence the barrier crossing probability of
the underlying asset. In this section, we investigate how jumps change barrier crossing
probabilities and option values.
2.1 Methodology
We use the down-and-out call option as an example to illustrate the e�ect of the jump risk
on barrier option pricing. The option ceases to exist when the underlying asset price S hits
or falls below a constant barrier H. We assume that the maturity of the option is T and
the exercise price of the option is X, where X > H. If the underlying asset price follows a
standard log-normal di�usion process, the option price can be expressed analytically as
DAOC(S;X;H; r; �; T ) = BSC(S;X; r; �; T ) ��H
S
� �1
BSC
H2
S;X; r; �; T
!; (6)
where DAOC is the down-and-out call price, BSC is the Black-Scholes standard European
call price, and = 2r=(�2). Other notation is de�ned as before.
We now consider the prices of down-and-out call options with jump di�usion processes
and compare them with the prices of the corresponding options with log-normal di�usion
processes obtained through equation (6).
Denote � as the �rst passage time for the underlying asset price S crossing the barrier
H, that is
� := infftjSt � H; t � 0g:
If � � T , the option is knocked-out before its maturity and becomes worthless. Otherwise,
the option is not knocked out during its life and receives a payment at its maturity just like
a standard European call. The �nal payo� to the option can be expressed as:
WT =
8><>:
0; if � � T ,
max(0; ST �X) otherwise:(7)
It follows immediately that
DAOC(S;X;H;T ) = exp(�rT )EQ[WT ]
5
= exp(�rT )EQ[max(0; ST �X)j� > T ]Q(� > T ); (8)
where Q(� > T ) is the probability that the �rst passage time � is bigger than the maturity
of option T under the equivalent martingale measure Q.
Theorem 1 Divide the time interval [0; T ] into n equal subperiods and de�ne
ti =i
nT;
= fS�ti > H; i = 1; 2; � � � ; ng;
where S�ti is de�ned recursively as
S�t0 = S;
ln(S�ti)� ln(S�ti�1) = si + yi � �i; i = 1; 2; � � � ; n:
Here si, yi, and �i are mutually and serially independent random variables drawn from
si � N((r � �2=2� ��)T=n; �2 � T=n);
�i � N(��; �2�);
and
yi =
8><>:
0; with prob. 1� � � T=n
1; with prob. � � T=n
Then the option price DAOC(S;X;H;T ) given in equation (8) can be expressed as
DAOC(S;X;H;T ) = exp(�rT ) limn!1
EQ[max(0; S�T �X)j]Q(): (9)
Brie y speaking, the theorem holds because in a very small time period, no more than
one jump can occur almost surely and the di�usion process cannot move a long distance.
The theorem can be easily extended to value options with time-varying barriers or multiple
barriers. The proof of the theorem is outlined in the appendix.
Theorem 1 provides a straightforward framework to price barrier options. The following
is a Monte Carlo valuation procedure based on this framework, which involves nothing more
complex than making random draws from some simple distributions.
6
� Step (1). Divide the time interval [0; T ] into n equal subperiods for a su�ciently large4
n and denote ti := T � i=n.
� Step (2). Do Monte Carlo replications for M (j = 1; 2; � � � ;M) times. For each j, run
the following sub-procedures for i = 1; � � � ; n.
a) Generate a series of mutually and serially independent random vectors (si; �i; yi)
as described in Theorem 1. Let S�t0 = X and calculate ln(S�ti) or S�
tiaccording to the
formula
ln(S�ti) = ln(S�ti�1) + si + yi � �i:
If S�ti � H, let Wj = 0 and go to the next j.
b) Let Wj = max(0; S�tn �X) and go to the next j.
� Step (3). Let C = exp(�rT )PMj=1Wj=M . C will be a numerical solution to the
option price.
2.2 Numerical Simulations
The following numerical illustrations show some rich implications of the model. In these
illustrations, the current stock price S is chosen as $20.00 and the barrier price H is $16.00.
Equation (5) implies that the volatility of underlying asset price growth d ln(S) is
�2S := Var(d ln(S))=dt = �2 + � � �2�; (10)
4The appropriate choice of n depends on the remaining maturity of the option T and the distance between
the current underlying price and the barrier j ln(S)�ln(H)j. If T and j ln(S)�ln(H)j are small, then to avoid
the problem that the barrier is `hit but missed' and to increase the pricing accuracy (see, e.g., Geman and
Yor 1996), the step size T=n should also be very small. Interestingly, we �nd that simulated option prices
are less sensitive to step sizes under jump processes than under di�usion processes. This is because when
the step size is small enough, it is unlikely that there is more than one jump in one step. Following Geman
and Yor (1996), we choose the step size as 1/4 day. It takes about one minute to value a one-month option
with a SPARC 20 computer if we perform 10,000 Monte Carlo paths. To improve the e�ciency of Monte
Carlo simulations, some numerical techniques may be used. See, for example, Chidambaran and Figlewski
(1995) for a detailed discussion.
7
if �� = 0. For convenience, �� will be set to zero in our numerical simulations. The value
of �2S will be kept as a constant so that the changes in option values are really from the
changes in the relative importance of the jump component rather than from the changes in
the overall volatility of the underlying asset price. We use the following data (expressed at
a monthly frequency) in simulations:
�S = 0:10: Monthly volatility of asset prices is 10 percent. This implies approximately
a 35 percent annual volatility.
� = 0:03: On average, there are 0.03 jumps in a month or approximately one jump
every three years. Jumps are rare events in our simulations.
�2� = 0:10. As argued by Merton, jumps are rare events which cause substantial changes
in asset prices. Assuming �2� = 0:10 implies approximately a 30 percent standard deviation
of the log change in the asset price caused by a jump. For rare (on average, once every three
years in this case) and substantial jumps, this number does not seem implausible. Certain
kind of growth stocks may have more volatile jumps. For comparison reason, �2� = 0:25 will
also be used in simulations.
The main purpose of our numerical simulations is to provide readers with some avor or
conceptual insight about the potential impact of the jump risk on option prices. These sim-
ulations are not substitutes for further empirical studies. A rigorous empirical veri�cation
of the model is helpful, and is best treated as a subject for a separate paper.
2.2.1 At-the-Money Call Options
Table 1 illustrates the e�ect of the jump risk on the price of at-the-money call options
with the volatility of the underlying asset price remaining constant. For down-and-out call
options, jumps generally lower the price of short maturity options, while increasing the
value of long maturity options. For example, the price of a 1/2-month at-the-money down-
and-out call option with �2� = 0:25 is about $0.365, but a similar option with the log-normal
di�usion process is valued at as much as $0.589. On the other hand, the price of a two-year
at-the-money down-and-out call with �2� = 0:25 is about $4.19, but a similar option with
8
a log-normal di�usion process is only worth $3.79. For standard at-the-money European
calls, the jump risk consistently lowers the option prices over various maturities, but such
e�ects become less important when option maturities are longer. The table suggests that
the jump risk may be far more important than the second order e�ects in the valuation
of path-dependent options. For barrier options, the e�ects of the jump risk option prices
do not disappear with the lengthening of option maturities. Ignoring the jump risk may
lead to large biases in option prices even if the volatilities of the underlying asset prices are
correctly estimated.
The �nding of Table 1 that the relation between option prices over short maturities is
reversed over longer maturities is surprising. Since this result is not found for the prices of
standard European options, it likely has something to do with knock-out probabilities. A
detailed analysis is in order.
Table 2 shows the knock-out probabilities of options with di�erent volatilities of the
jump amplitude, �2�. The table illustrates a pattern of knock-out probabilities which roughly
matches the pattern of the prices of down-and-out call options: Holding constant the volatil-
ity of the underlying asset price S, over short maturities, an option with a more volatile
jump component is more likely to be knocked-out than is an option with a more volatile
di�usion component; over longer maturities, the relation is reversed: An increase in the
jump volatility �� can substantially reduce the knock-out probability of long-term options.
Consider a two year knock-out call as an example. Under our parametric speci�cation,
when �� = 0, the knock-out probability is 0.65, but when �� = 0:25, this probability is only
0.45.
Because a di�usion component is much less likely to produce a large change in the asset
price S over a short period, a short-maturity option with a more volatile jump component
is more likely to be knocked out than is a same maturity option with a less volatile jump
component. What is interesting here is that for long maturity options, a high jump volatility
is generally associated with a low knock-out probability. We now outline some intuition for
this result.
9
For a given T > 0 which is not very small in magnitude, an increase in the volatility
of a di�usion process can substantially increase the knock-out probability during [0; T ].
However, for a jump process, the e�ect of the jump size volatility �� on the knock-out
probability is largely limited by the jump intensity �. If � is very small such that �T is also
small, the probability that there is at least one jump in period [0; T ] is approximately �T .
As a result, no matter how large the jump size volatility �2� is, the knock-out probability
in period [0; T ] caused by the jump process is always smaller than �T , even though �T is
already small. In this case, an increase in �2� mainly a�ects the value of asset at knock-out
time but has a very small e�ect on the knock-out probability. This intuition is made more
rigorous in the following example.
Consider two extreme S processes for illustration. The �rst one is a pure di�usion
process with the volatility �2 and the second one is a pure jump process with a small
jump intensity � and a large volatility of jump amplitude �2� := Var(ln(�)). I assume that
�2 = � � �2� = 0:01, which is the same as the volatility of ln(S) used in Table 2.
Denote D(T ) as the cumulative distribution function of the �rst passage time to the
barrier H for the pure di�usion process and denote J(T ) as the cumulative distribution
function of the �rst passage time to the barrier for the pure jump process, where T is the
maturity time. Using the result of Harrison (1990), we have
D(T ) = N
� ln(S=H) + (r � �2=2)T
�pT
!
+(S=H)(1� 2r
�2)N
� ln(S=H)� (r � �2=2)T
�pT
!: (11)
Assuming that S = 20 at time zero and H = 16 as in Table 2, we obtain immediately
that D(1=4) = 0:00001 and D(18) = 0:599.
There is no explicit expression for J(T ). However, in a pure jump process with a
positive drift, crossing the barrier must be caused by jumps. Assume that � = 0:03 and
that �2� = 0:01=0:03 = 1=3. If T = 1=4 (one month), then the probability of one jump
in [0; T ] is about �T = 0:0075 and the probability of two or more jumps in [0; T ] is small
enough to ignore. If a jump occurs at time t < 1=4 and there are no other jumps before t,
10
the probability that St falls to or below H is
N(� ln(S=H)=��) = N(� ln(20=16)=q1=3) = 0:35:
As a result, we have J(1=4) � 0:0075 � 0:35 = 0:003.
Now let's consider T = 18. The probability that there is no jump in the time interval
[0; T ] is
exp(��T ) = exp(�0:03 � 18) = 0:583:
That is, the probability that there are one or more jumps in [0; T ] is 1 � 0:583 = 0:417.
Denote d < 1 as the conditional probability of crossing the barrier if there are jumps. Then
we have
J(18) = 0:417 � d� 0:599:
From the above examples, we see that J(1=4)� D(1=4) and that J(18)� D(18). That
is, a jump process is more likely than is a di�usion process to cause a knock-out over a short
horizon but less likely to cause a knock-out over a long horizon.
2.2.2 In-the-Money and Out-of-the-Money Options
Tables 3 and 4 report the example prices of the in-the-money and the out-of-the-money call
options.
Let's discuss in-the-money options �rst. For options with very short maturities, knock-
out probabilities are very low so that the prices of the standard options and the correspond-
ing barrier options are very close.
As maturities become longer, the prices of the standard options and the barrier options
display di�erent patterns. Similar to the at-the-money cases shown in Table 1, the values
of the in-the-money standard European call options with jumps tend to be lower than
those of the corresponding maturity options with continuous di�usion processes. Holding
constant the volatility of the underlying asset price, the prices of options become insensitive
to the jump risk when option maturities are su�ciently long. For the barrier options, jumps
reduce the option values in the medium range of maturities, same as that which happens to
11
the standard in-the-money call options. This is because with these maturities, the knock-
out probabilities have not become large enough to change the properties of barrier options
inherited from their standard counterparts. As the maturities become long enough, the
knock-out probabilities of options with di�usion processes become much larger than the
knock-out probabilities of options with jumps, holding �2S constant (See Table 2). As a
result, the options with more volatile jump components become more valuable.
We now consider out-of-the-money calls.
For both the standard and the down-and-out out-of-the-money calls, jumps generally
raise the value of options near expiration. This is because the underlying asset prices with
more volatile jump components are more likely to move a su�cient distance to pass the
strike price X (X > S) in a short period, as discussed earlier.
As maturities get longer, the standard out-of-the-money calls with jumps become less
valuable, because as was discussed previously, the di�usion processes are more likely to
pass a threshold level (X in this situation) in a relatively long period. For down-and-out
calls, the situation is more complicated. Jumps have two opposite e�ects on out-of-the-
money option values. On the one hand, similar to that which occurs with the standard
out-of-the-money calls, jumps tend to lower option values by reducing the probability that
the underlying asset will cross the strike price X. On the other hand, jumps may increase
option values by reducing knock-out probabilities, as shown in Table 2. The net impact
depends on the relative importance of these two e�ects. In the example shown in Table 4,
the �rst e�ect dominates over medium horizons and the second e�ect is more important
over long horizons. As a result, jumps reduce the prices of down-and-out out-of-the-money
calls with medium maturities but increase the prices of options with long maturities.
12
3 Pricing Lookback Options
Lookback options give the holder the right, at maturity, to exercise the option at the
most favorable price reached by the underlying asset during the life of the option.5Take
a lookback call option as an example. If Smin is the minimum price reached and ST is
the asset price at the maturity of the option, T , the payo� from the option is ST � Smin.
According to Lemma 1, the price of this option is exp(�rT )EQ[ST � Smin].
Theorem 2 Using the same approach as in Theorem 1, the price of a lookback option can
be expressed as
LB(S; SM ;T ) = exp(�rT ) limn!1
EQ[S�T �min(SM ;S�t0 ; S
�
t1; � � � ; S�tn)]; (12)
where SM is the minimum asset price reached to date and all other notation is de�ned as
in Theorem 1.
The proof of this theorem is similar to that of Theorem 1. Based on this theorem, a
numerical approach similar to the one used to value barrier options can be used to value
lookback options.
Table 5 presents a numerical example to illustrate the e�ects of jumps on lookbacks.
In this table, the volatility of stock price �2S is �xed at a constant level 0:01. The Table
shows that jumps generally reduce the price of lookback call options. For short maturity
lookbacks, jumps have a relatively small impact on option prices, but when option maturities
get relatively long, jumps will a�ect option prices signi�cantly. More speci�cally, in the
numerical example shown here, a 1/2-month lookback call with log-normal process is worth
about $2.12. With a jump-di�usion process, such an option is valued at $2.08 if �2� = 0:10
and at $2.06 �2� = 0:25. If the maturity of a lookback call becomes one year, its price is
about $5.64 with a log-normal process and becomes as low as $4.74 when jumps are present
(� = 0:03 and �2� = 0:25).
5For detailed discussions of lookback options, see, for example, Conze and Viswanathan (1991) and
Goldman, Sosin, and Gatto (1979).
13
4 Conclusions
This paper presents a general framework for valuing a variety of path-dependent exotic
options in the presence of jumps in the underlying price process. It shows that ignoring
jumps may lead to seriously biased options prices. For instance, for at-the-money down-
and-out barrier calls, ignoring jumps will lead to upward biases in option prices with short
maturities, but downward biases in option prices with long maturities. For lookback call
options, ignoring jumps will often lead to upward biases in options prices, especially for
those with relatively long maturities. The biases are generally signi�cant in magnitude.
5 Appendix
This appendix outlines the proof of Theorem 1.
Proof of Theorem 1: Equation (5) implies that
ln(Sti)� ln(Sti�1) = si +
�iXj=0
�ij; (13)
where
si � N((r � �2=2� ��)T=n; �2 � T=n);
�ij � N(��; �2�);
and
�i = k; with prob. �k � exp(���T=n)k! � (T=n)k, k = 0; 1; � � � :
Since Prob(�i = k) = o(T=n) for k > 1, by the de�nitions of S�ti and , we have
DAOC(S;X;H;T ) = exp(�rT )EQ[max(0; ST �X)j� > T ]Q(� > T )
= limn!1
exp(�rT )EQ[max(0; ST �X)jSti > H; i = 1; � � � ; n]
�Q(Sti > H; i = 1; � � � ; n)
= limn!1
exp(�rT )EQ[max(0; S�T �X)jS�ti > H; i = 1; � � � ; n]
�Q(S�ti > H; i = 1; � � � ; n)
= limn!1
exp(�rT )EQ[max(0; S�T �X)j]Q():
14
References
[1] Amin, K.I.(1993): \Jump di�usion option valuation in discrete time," Journal of Fi-
nance 48, 1833-1863.
[2] Bates, D.S.(1991): \The crash of '87: Was it expected? The evidence from options
markets," Journal of Finance 46, 1009-1044.
[3] Bates, D.S.(1996): \Jumps and stochastic volatility: Exchange rate processes implicit
in Deutsche Mark options," Review of Financial Studies 9, 69-107.
[4] Ball, C. and W. Torous(1985): \On jumps in common stock prices and their impact
on call option pricing," Journal of Finance 40, 155-173.
[5] Black, F.(1975): \Fact and fantasy in the use of options," Financial Analysts Journal
31 (July/August 1975), 36-72.
[6] Boyle, P.P. and S.H. Lau(1994): \Bumping up against the barrier with the binomial
method," Journal of Derivatives, Summer 1994, 6-14.
[7] Chidambaran, N.K. and S. Figlewski(1995): \Streamlining Monte Carlo simulation
with the quasi-analytic method: Analysis of a path-dependent option strategy," Journal
of Derivatives, Winter 1995, 29-52.
[8] Conze, A. and Viswanathan(1991): \Path dependent options: The case of lookback
options," Journal of Finance 46, 1893-1907.
[9] Cox, J.C. and S.A. Ross(1976): \The valuation of options for alternative stochastic
processes," Journal of Financial Economics 3, 145-166.
[10] Das, S.R., S. Foresi, and R.K. Sundaram(1996): \Modeling asset price processes in
�nance," preprint.
[11] Geman, H. and M. Yor(1996): \Pricing and hedging double-barrier options: A proba-
bilistic approach," Mathematical Finance 6, 365-378.
15
[12] Goldman, M.B., H.B. Sosin, and M.A. Gatto(1979): \Path dependent options: `Buy
at the low, sell at the high'," Journal of Finance 34, 1111-1127.
[13] Harrison, J.M.(1990): Brownian motion and stochastic ow systems, Krieger Publish-
ing Company: Florida.
[14] Heynen, H. and H. Kat(1994): Crossing barriers, Risk, June 1994, 46-51.
[15] Jarrow, R.A. and A. Rudd(1983): Option Pricing, Irwin.
[16] Jorion, P.(1988): \On jump processes in the foreign exchange and stock markets,"
Review of Financial Studies 1, 427-445.
[17] Kon, S.J.(1984): \Models of stock returns|A comparison," Journal of Finance 39,
147-165.
[18] Merton, R.C.(1976a): \Option pricing when underlying stock returns are discontinu-
ous," Journal of Financial Economics 3, 125-144.
[19] Merton, R.C.(1976b): \The impact on option pricing of speci�cation error in the un-
derlying stock price returns," Journal of Finance 31, 333-350.
[20] Rich, D.R.(1994): \The mathematical foundations of barrier option-pricing theory,"
Advances in Futures and Options Research 7, 267-312.
[21] Ritchken, P.(1995): \On pricing barrier options," Journal of Derivatives, Winter 1995,
19-28.
[22] Rubinstein, M. and E. Reiner(1991): Breaking down the barriers, Risk, September
1991, 28-35.
16
Table 1: Prices of At-the-Money Call Options (Monthly Frenquency). S = 20,
X = 20, � = 0:03, �� = 0, �S = 0:1, �2 = �2S � ��2�.
Maturity 1/4 1/2 1 2 3 6 9 12 18 24
Down-and-Out Call Options (H = 16)
�2� = 0:00 0.411 0.589 0.846 1.223 1.518 2.160 2.603 2.937 3.430 3.790
�2� = 0:10 0.362 0.522 0.765 1.141 1.431 2.115 2.580 2.958 3.527 3.939
�2� = 0:25 0.242 0.365 0.558 0.869 1.140 1.807 2.376 2.840 3.573 4.187
Standard European Call Options
�2� = 0:00 0.411 0.589 0.846 1.224 1.524 2.231 2.798 3.292 4.147 4.889
�2� = 0:10 0.363 0.529 0.776 1.147 1.446 2.160 2.735 3.236 4.103 4.853
�2� = 0:25 0.243 0.367 0.565 0.883 1.155 1.842 2.422 2.939 3.845 4.635
Table 2: Knock-out Probabilities of Down-and-Out Call Options (Monthly Fren-
quency). S = 20, H = 16, � = 0:03, �� = 0, �S = 0:1, �2 = �2S � ��2�.
Maturity 1/4 1/2 1 2 3 6 9 12 18 24
�2� = 0:00 0.000 0.002 0.026 0.115 0.198 0.362 0.457 0.519 0.599 0.649
�2� = 0:25 0.002 0.004 0.014 0.066 0.133 0.289 0.388 0.455 0.547 0.603
�2� = 0:50 0.003 0.005 0.011 0.023 0.041 0.120 0.203 0.273 0.378 0.451
17
Table 3: Prices of In-the-Money Call Options (Monthly Frenquency). S = 20,
X = 17:5, � = 0:03, �� = 0, �S = 0:1, �2 = �2S � ��2�.
Maturity 1/4 1/2 1 2 3 6 9 12 18 24
Down-and-Out Call Options (H = 16)
�2� = 0:00 2.523 2.557 2.657 2.875 3.064 3.474 3.754 3.966 4.287 4.531
�2� = 0:10 2.527 2.563 2.642 2.849 3.045 3.517 3.828 4.091 4.478 4.761
�2� = 0:25 2.535 2.570 2.640 2.793 2.942 3.419 3.863 4.208 4.776 5.271
Standard European Call Options
�2� = 0:00 2.529 2.557 2.658 2.887 3.109 3.695 4.197 4.644 5.429 6.115
�2� = 0:10 2.529 2.564 2.652 2.861 3.071 3.647 4.149 4.598 5.389 6.082
�2� = 0:25 2.536 2.572 2.644 2.799 2.962 3.459 3.932 4.373 5.175 5.890
Table 4: Prices of Out-of-the-Money Call Options (Monthly Frenquency). S = 20,
X = 23, � = 0:03, �� = 0, �S = 0:1, �2 = �2S � ��2�.
Maturity 1/4 1/2 1 2 3 6 9 12 18 24
Down-and-Out Call Options (H = 16)
�2� = 0:00 0.001 0.015 0.088 0.293 0.507 1.088 1.561 1.949 2.549 3.003
�2� = 0:10 0.014 0.032 0.095 0.280 0.478 1.046 1.515 1.925 2.585 3.086
�2� = 0:25 0.032 0.064 0.124 0.269 0.418 0.910 1.398 1.820 2.551 3.212
Standard European Call Options
�2� = 0:00 0.001 0.015 0.088 0.293 0.508 1.105 1.632 2.109 2.959 3.712
�2� = 0:10 0.014 0.033 0.096 0.282 0.483 1.066 1.591 2.070 2.925 3.684
�2� = 0:25 0.033 0.065 0.131 0.275 0.433 0.933 1.424 1.891 2.753 3.532
18
Table 5: Prices of Lookback Call Options (Monthly Frenquency). S = 20, SM =
18, � = 0:1, �� = 0, �2S = 0:252, �2 = �2S � ��2�.
Maturity (yrs) 1/4 1/2 1 2 3 6 9 12
�2� = 0:00 2.034 2.119 2.347 2.802 3.206 4.192 4.977 5.642
�2� = 0:10 2.020 2.084 2.243 2.682 3.024 3.956 4.625 5.291
�2� = 0:25 2.006 2.061 2.154 2.410 2.651 3.358 4.103 4.738
19